Feature Extraction And Recognition Based On The Lower Leg And Foot

Biometric identification technology is a kind of personal identification technology,and people focus on the security day by day, so we make this technology in-depth study. On the long distance, because of fingerprint and face biological characteristics’s perception, It is unable to use, gait make up this shortcoming, therefore, gait recognition has a clear advantage in visual surveillance. First, we pretreatment of gait. Through analyze and compare the more commonly used motion detection method, we choose background subtraction method to achieve the gait sequence detection; we analyze the gait cyclical movement and use the toe trajectory to calculate the cycle information. Then, we put forward gait feature recognition method of toe trajectory based on the view of "the human low body provides the more gait information "[2] For each gait sequence, we manually extract the toe coordinates, then get a toe motion sequences, and analyze the normalized toe trajectory by wavelet packet analysis, extract the signal energy in different frequency bands as a feature vector; and then, use SupportVector Machine (SVM) to achieve gait classification. Finally, we use the feature fusion method to effectively solve incomplete characterization of single characteristical problems in matching decision integrating the lower leg joint features. In the process of integration, it gives full play to their complementarity of gait characteristics by using Dempster-Shafer evidence theory (DS Evidence Theory). Simulation results on CASIA database show that the recognition rate of this method is 95%, and obviously higher than the recognition method of only using single feature, it shows that this is an effective identification method.